Assessment of Coal Mine Emergency Response Capability Based on Bayesian Network

Abstract

Coal accounts for over 53% of China's primary energy consumption, making its production safety a critical pillar of national energy security. However, major coal mine accidents persist, largely attributable to inadequate emergency response capabilities. To address the challenges of information uncertainty and complex indicator interdependencies, this paper constructs a Bayesian Network (BN)-based evaluation model for coal mine emergency response capability. Following the Coal Mine Safety Regulations, an index system encompassing 4 dimensions—personnel, equipment, environment, and management—with 14 specific indicators was established. The BN structure, comprising 19 nodes and 21 directed edges, was built using GeNIe software, with Conditional Probability Tables parameterized via a logic-grading method based on expert-defined rules and indicator grading standards. A case study at a large Shandong mine demonstrates that the comprehensive capability is rated "Excellent" (probability 38%), with a cumulative probability of 70% for "Excellent" and "Good", consistent with the mine's actual safety status during its intelligent transformation. Sensitivity analysis identifies "safety violation inspections" as the most influential factor.